[Tzdist] Fun with tzdata and non-Gregorian DST rules

[Ken Murchison <[e-mail]>]

All,

I found this interesting/amusing so I thought I'd share it.

I was poking around tzdata trying to figure out why Asia/Tehran compiled 
with vzic doesn't include a POSIX tz string and I found out that its 
because the Iranian DST rules are based on the Persian calendar, not the 
Gregorian calendar.  This is why the Rule for Iran has a list of changes 
in groups of 1-3 years up to 2038 and no Rule that extends out into the 
future.

As can be expected, an iCalendar representation for Asia/Tehran based on 
tzdata also has a list of dates and no recurrence rule that repeats forever.

However, a much more compact iCalendar representation for this zone that 
does extend to eternity can be created using Non-Gregoran Recurrence 
Rules in iCalendar <https://tools.ietf.org/html/rfc7529> (RSCALE).

I created a new zone named Asia/Tehran_Persian using the RSCALE 
extension and installed it on my tzdist test server.

For those that are interested, you can look at the 2 zones here:

https://cyrus-test.andrew.cmu.edu/tzdist/zones/Asia/Tehran
https://cyrus-test.andrew.cmu.edu/tzdist/zones/Asia/Tehran_Persian

If you want to compare the observances, especially those from 1991-2038 
(yes, they match), you can view them here in JSON format:

https://cyrus-test.andrew.cmu.edu/tzdist/zones/Asia/Tehran/observances?start=0000-01-01T00:00:00Z&end=2500-01-01T00:00:00Z
https://cyrus-test.andrew.cmu.edu/tzdist/zones/Asia/Tehran_Persian/observances?start=0000-01-01T00:00:00Z&end=2500-01-01T00:00:00Z

and here in "zdump -V" format:

https://cyrus-test.andrew.cmu.edu/tzdist/zones/Asia/Tehran/observances?start=0000-01-01T00:00:00Z&end=2500-01-01T00:00:00Z&format=application/zdump
https://cyrus-test.andrew.cmu.edu/tzdist/zones/Asia/Tehran_Persian/observances?start=0000-01-01T00:00:00Z&end=2500-01-01T00:00:00Z&format=application/zdump

For those that are really ambitious, you can look at a binary "tzfile" 
version of Asia/Tehran_Persian (which gets created from the iCalendar 
data) here:

https://cyrus-test.andrew.cmu.edu/tzdist/zones/Asia/Tehran_Persian?format=application/tzfile

This file is binary compatible with one created by vzic except I don't 
add a transition at 2440-01-01 like vzic does.  (I'm not even sure why 
that transition is there or what it means).


I briefly looked for other zones based on lunisolar calendars and didn't 
come up with any.  The Brazilian rules are essentially based around 
Easter and there is no way that I have found to represent Easter as a 
recurrence rule in iCalendar.

All of this is just for fun at this point because I don't believe that 
any calendaring clients have support for RSCALE yet.

-- 
Kenneth Murchison
Principal Systems Software Engineer
Carnegie Mellon University